Intelligent operation management method and system based on multi-device cooperation

Abstract

This application discloses a smart operation management method and system based on multi-device collaboration, relating to the field of multi-device collaborative control technology. The method includes: receiving real-time operating data from multiple heterogeneous devices, fusing the real-time operating data to generate comprehensive status information, inputting the comprehensive status information into a pre-trained collaborative decision model, outputting a target collaborative strategy matching the current operating scenario, determining a target device cluster that needs to perform collaborative tasks based on the target collaborative strategy and the comprehensive status information, generating differentiated control instructions for each device in the target device cluster according to the objective of the collaborative task, and distributing the differentiated control instructions to the corresponding target devices to drive the target device cluster to collaboratively complete the collaborative task. This approach overcomes the technical problem of low collaborative efficiency caused by rigid decision-making during multi-heterogeneous device collaborative operation.

Description

Technical Field

[0001] This application relates to the field of multi-device collaborative control technology, and in particular to a smart operation management method and system based on multi-device collaboration. Background Technology

[0002] Existing multi-device collaborative management technologies primarily rely on centralized control systems with pre-defined rules or distributed coordination schemes based on fixed algorithms. Specifically, system administrators or algorithm engineers pre-define a series of "if-then" collaborative logics, task allocation strategies, and execution flows based on historical experience or abstractions of limited scenarios. For example, when sensor A detects an object, it calls robot B to perform the operation, specifying its route and action sequence. Since pre-defined rules are inherently static, their decision-making logic cannot cover all possible unforeseen situations in the real world, such as temporarily inserted high-priority tasks, sudden equipment failures, or dynamic obstacles in the environment. When encountering "long-tail scenarios" not defined in the rule base, the system either fails to make effective decisions or makes suboptimal or even incorrect decisions, leading to task stagnation or execution errors.

[0003] The above content is only used to help understand the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention

[0004] The main purpose of this application is to provide a smart operation management method and system based on multi-device collaboration, which aims to solve the technical problem of rigid decision-making and low collaboration efficiency when multiple heterogeneous devices are operated collaboratively in the prior art.

[0005] To achieve the above objectives, this application provides a smart operation management method based on multi-device collaboration, the method comprising:

[0006] Receive real-time operating data from multiple heterogeneous devices, and perform fusion processing on the real-time operating data to generate comprehensive status information;

[0007] The comprehensive state information is input into a pre-trained collaborative decision-making model, which outputs a target collaborative strategy that matches the current operational scenario. The collaborative decision-making model is built based on deep reinforcement learning.

[0008] Based on the target collaboration strategy and the comprehensive status information, the target device cluster that needs to perform collaborative tasks is determined;

[0009] Based on the objective of the collaborative task, generate differentiated control instructions for each device in the target device cluster;

[0010] The differentiated control commands are distributed to the corresponding target devices to drive the target device cluster to collaboratively complete the collaborative task.

[0011] In one embodiment, the step of receiving real-time operating data from multiple heterogeneous devices and fusing the real-time operating data to generate comprehensive status information includes:

[0012] Receive real-time operating data from multiple heterogeneous devices, perform data cleaning and format standardization on the received real-time operating data from each device, and obtain standardized data;

[0013] Based on a multi-source information fusion algorithm, data from different devices but describing the same operating object in the standardized data are correlated and complemented to obtain correlated data;

[0014] A dynamic digital twin model describing the operating environment is constructed from the associated data. The associated data is then fused and processed based on the dynamic digital twin model to generate comprehensive status information.

[0015] In one embodiment, after the steps of constructing a dynamic digital twin model describing the operating environment based on the associated data, and fusing the associated data according to the dynamic digital twin model to generate comprehensive status information, the method further includes:

[0016] A probabilistic model is established for the key physical parameters in the dynamic digital twin model, and sensor data is acquired in real time. The key physical parameters are calibrated based on the sensor data to obtain calibration parameters.

[0017] The historical data within a preset time window is compared with the corresponding comprehensive status to determine the model confidence of the probability model;

[0018] When the model confidence level is lower than the confidence threshold, the dynamic digital twin model is reconstructed based on the calibration parameters.

[0019] In one embodiment, the step of inputting the comprehensive state information into a pre-trained collaborative decision-making model and outputting a target collaborative strategy matching the current operational scenario includes:

[0020] Define a reward function that aims to optimize task completion efficiency and system resources;

[0021] The comprehensive state information and reward function are input into the pre-trained collaborative decision-making model, so that the pre-trained collaborative decision-making model performs cumulative reward evaluation on collaborative decisions based on the comprehensive state information and the reward function, and obtains a reward evaluation value;

[0022] A probability distribution of actions is generated based on the reward evaluation value, and the target action sequence is determined by the action sequence of the probability distribution of the action sequence corresponding to the maximum reward evaluation value.

[0023] The target action sequence is used as a target collaboration strategy that matches the current operational scenario.

[0024] In one embodiment, after the step of using the target action sequence as a target collaborative strategy matching the current operational scenario, the method further includes:

[0025] The target cooperative strategy was simulated and verified, and the simulation verification results were obtained.

[0026] If the simulation verification result shows that the execution effect of the target collaborative strategy does not reach the predetermined threshold, the verification is deemed to have failed, and an abnormal scenario is identified.

[0027] The pre-trained collaborative decision-making model is retrained based on the abnormal scenario to obtain an optimized collaborative decision-making model;

[0028] The optimized collaborative decision-making model is then released in a canary release phase, replacing the pre-trained collaborative decision-making model.

[0029] In one embodiment, the step of determining the target device cluster that needs to perform collaborative tasks based on the target collaboration strategy and the comprehensive state information includes:

[0030] Analyze the target collaboration strategy to obtain the core task type and required equipment capability combination corresponding to the target collaboration strategy;

[0031] Based on the combination of device capabilities, a set of candidate devices that meet the criteria is selected from the currently online and available devices;

[0032] Based on the comprehensive status information, the real-time load and geographical location information of the candidate device set are determined, the collaborative communication cost is evaluated, and the target device cluster that needs to perform collaborative tasks is determined based on the real-time load, the geographical location information, and the evaluated collaborative communication cost.

[0033] In one embodiment, the step of generating differentiated control instructions for each device in the target device cluster based on the objective of the collaborative task includes:

[0034] The collaborative task is decomposed into multiple atomic subtasks with logical order and dependencies.

[0035] Based on the roles and capabilities of each device in the target device cluster, the atomic subtasks are assigned to the corresponding devices;

[0036] For each assigned atomic subtask, a precise start timestamp and timeout threshold are set, and the interaction interface protocol required for data transfer between devices is configured to form the differentiated control commands.

[0037] In one embodiment, after the step of distributing the differentiated control commands to the corresponding target devices to drive the target device cluster to collaboratively complete the collaborative task, the method further includes:

[0038] It receives instruction execution feedback data and new environmental status data from the target device cluster in real time.

[0039] The feedback data is compared with the expected execution status. If there is equipment malfunction, execution delay or sudden environmental change, it is determined to be a coordination deviation.

[0040] Based on the type and severity of the coordination deviation, the unexecuted control instructions are dynamically adjusted or the coordination decision-making model is triggered to make a new decision.

[0041] In one embodiment, the step of dynamically adjusting unexecuted control instructions or triggering the collaborative decision-making model to make a re-decision based on the type and severity of the coordination deviation includes:

[0042] If a device fails, the failed device is isolated from the target device cluster, and the unfinished atomic subtasks of the failed device are reassigned to other available devices in the cluster.

[0043] If a delay in execution timing is detected, the delay amount is calculated, and the start time of the unexecuted control instructions is proportionally compressed or delayed based on the delay amount.

[0044] All dynamic adjustment actions, reasons for adjustment, and results of adjustment are recorded in a log, and the collaborative decision-making model is triggered to make a new decision based on the log.

[0045] Furthermore, to achieve the above objectives, this application also proposes a smart operation management system based on multi-device collaboration, which includes:

[0046] The data fusion module is used to receive real-time operating data from multiple heterogeneous devices, and to fuse the real-time operating data to generate comprehensive status information.

[0047] The intelligent collaboration module is used to input the comprehensive state information into a pre-trained collaborative decision-making model and output a target collaboration strategy that matches the current operation scenario. The collaborative decision-making model is built based on deep reinforcement learning.

[0048] The device determination module is used to determine the target device cluster that needs to perform the collaborative task based on the target collaboration strategy and the comprehensive status information.

[0049] The instruction generation module is used to generate differentiated control instructions for each device in the target device cluster according to the objective of the collaborative task.

[0050] The dynamic control module is used to distribute the differentiated control commands to the corresponding target devices to drive the target device cluster to collaboratively complete the collaborative task.

[0051] In addition, to achieve the above objectives, this application also proposes a smart operation management device based on multi-device collaboration. The smart operation management device based on multi-device collaboration includes: a memory, a processor, and a computer program stored in the memory and executable on the processor. The computer program is configured to implement the steps of the smart operation management method based on multi-device collaboration as described above.

[0052] In addition, to achieve the above objectives, the present invention also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the intelligent operation management method based on multi-device collaboration as described above.

[0053] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the intelligent operation management method based on multi-device collaboration as described above.

[0054] This application provides a smart operation management method based on multi-device collaboration. It receives real-time operational data from multiple heterogeneous devices, fuses and processes this data to generate comprehensive status information, inputs this comprehensive status information into a pre-trained collaborative decision-making model, and outputs a target collaborative strategy matching the current operational scenario. Based on the target collaborative strategy and the comprehensive status information, it determines the target device cluster that needs to perform collaborative tasks. According to the objective of the collaborative task, it generates differentiated control instructions for each device in the target device cluster and distributes these differentiated control instructions to the corresponding target devices to drive the target device cluster to collaboratively complete the collaborative task. This method overcomes the technical problem of low collaborative efficiency caused by rigid decision-making during multi-heterogeneous device collaborative operation. Attached Figure Description

[0055] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0056] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0057] Figure 1 This is a flowchart illustrating an embodiment of the intelligent operation management method based on multi-device collaboration of this application;

[0058] Figure 2 This is a schematic diagram of the grayscale release of an embodiment of the intelligent operation management method based on multi-device collaboration in this application;

[0059] Figure 3 This is a schematic diagram of the module structure of the intelligent operation management system based on multi-device collaboration in an embodiment of this application;

[0060] Figure 4 This is a schematic diagram of the hardware operating environment involved in the intelligent operation management method based on multi-device collaboration in the embodiments of this application.

[0061] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0062] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0063] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0064] The main solution of this application embodiment is as follows: receiving real-time operating data from multiple heterogeneous devices and performing fusion processing on the real-time operating data to generate comprehensive status information; inputting the comprehensive status information into a pre-trained collaborative decision-making model and outputting a target collaborative strategy matching the current operating scenario, wherein the collaborative decision-making model is constructed based on deep reinforcement learning; determining the target device cluster that needs to perform collaborative tasks based on the target collaborative strategy and the comprehensive status information; generating differentiated control instructions for each device in the target device cluster according to the goal of the collaborative task; and distributing the differentiated control instructions to the corresponding target devices to drive the target device cluster to collaboratively complete the collaborative task.

[0065] Currently, existing multi-device collaborative management technologies primarily rely on centralized control systems with pre-defined rules or distributed coordination schemes based on fixed algorithms. Specifically, system administrators or algorithm engineers pre-define a series of "if-then" collaborative logics, task allocation strategies, and execution flows based on historical experience or abstractions of limited scenarios. For example, when sensor A detects an object, it calls robot B to perform the operation, specifying its route and action sequence. Since pre-defined rules are inherently static, their decision-making logic cannot cover all possible unforeseen situations in the real world, such as temporarily inserted high-priority tasks, sudden equipment failures, or dynamic obstacles in the environment. When encountering "long-tail scenarios" not defined in the rule base, the system either fails to make effective decisions or makes suboptimal or even incorrect decisions, leading to task stagnation or execution errors.

[0066] This application provides a solution that receives real-time operational data from multiple heterogeneous devices, fuses and processes this data to generate comprehensive status information, inputs this comprehensive status information into a pre-trained collaborative decision-making model, outputs a target collaborative strategy matching the current operational scenario, determines the target device cluster that needs to perform collaborative tasks based on the target collaborative strategy and the comprehensive status information, generates differentiated control instructions for each device in the target device cluster according to the objective of the collaborative task, and distributes these differentiated control instructions to the corresponding target devices to drive the target device cluster to collaboratively complete the collaborative task. This approach overcomes the technical problem of low collaborative efficiency caused by rigid decision-making during multi-heterogeneous device collaborative operation.

[0067] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device capable of performing the above functions, or a smart operation management device based on multi-device collaboration, etc. This embodiment does not specifically limit it. The following uses a smart operation management device based on multi-device collaboration as an example to describe this embodiment and the following embodiments.

[0068] All actions involving the acquisition of signals, information, or data in this application are carried out in accordance with the relevant data protection laws and policies of the country where the application is located, and with the authorization of the owner of the relevant device.

[0069] This application provides a smart operation management method based on multi-device collaboration, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the intelligent operation management method based on multi-device collaboration in this application.

[0070] In this embodiment, the intelligent operation management method based on multi-device collaboration includes steps S10 to S50:

[0071] Step S10: Receive real-time operating data from multiple heterogeneous devices, and perform fusion processing on the real-time operating data to generate comprehensive status information.

[0072] It should be noted that heterogeneous devices refer to different types of devices that differ in hardware architecture, operating system, communication protocol, function, or data format. Real-time operational data refers to time-sensitive data streams continuously or periodically generated by field devices during operation, such as sensor readings, device status, actuator feedback, and video surveillance streams. Comprehensive status information refers to the output result after data fusion processing.

[0073] Understandably, a set of adapters or edge gateways establishes connections with various heterogeneous devices, receives their raw data streams, and immediately performs data cleaning and format standardization, transforming heterogeneous data into homogeneous data. Subsequently, multi-source information fusion algorithms are applied to perform spatiotemporal alignment, data correlation, and complementarity verification on this standardized data. For example, location information from an RFID readout is fused with location information from a visual recognition system to obtain a more accurate location estimate. Finally, based on this fused data, a dynamic digital twin model is built or updated in the system. This model represents the generated comprehensive state information, providing a real-time, reliable, and global contextual environment for subsequent decision-making modules.

[0074] In one feasible implementation, the step of receiving real-time operating data from multiple heterogeneous devices and fusing the real-time operating data to generate comprehensive status information includes:

[0075] Receive real-time operating data from multiple heterogeneous devices, perform data cleaning and format standardization on the received real-time operating data from each device, and obtain standardized data;

[0076] Based on a multi-source information fusion algorithm, data from different devices but describing the same operating object in the standardized data are correlated and complemented to obtain correlated data;

[0077] A dynamic digital twin model describing the operating environment is constructed from the associated data. The associated data is then fused and processed based on the dynamic digital twin model to generate comprehensive status information.

[0078] It should be noted that data cleaning refers to identifying and processing errors, anomalies, and missing values ​​in raw data. This includes filtering out sensor readings that are clearly outside the reasonable range, or using interpolation to fill in temporary data loss, thus ensuring data quality. An operating object refers to an entity monitored and controlled within the operational environment; it can be a physical entity or a logical entity. A dynamic digital twin model refers to a virtualized, dynamically updated, and highly faithfully synchronized digital representation of the physical operating environment and its internal entities (equipment, objects, tasks). It is not merely a static 3D model, but a living model that can dynamically evolve based on real-time data and contains physical rules and logic; it is a carrier of comprehensive state information.

[0079] In its implementation, upon receiving real-time operational data from multiple heterogeneous devices, the system performs data cleaning and format standardization on the received data. It pre-defines valid range rules for each data type, discarding or smoothing data that exceeds these ranges. Simultaneously, a specific parser is written for each device protocol, mapping its data content to standardized internal data model fields. Then, Kalman filtering is used to fuse observations of the same state from different sensors; this process involves two steps: prediction and update. The specific formula is as follows:

[0080] Prediction phase:

[0081]

[0082]

[0083] Update phase:

[0084]

[0085]

[0086]

[0087]

[0088]

[0089] Where k is the current time step, x is the system state vector, z is the observation vector, F is the state transition matrix, describing how the state evolves over time, H is the observation matrix, describing how the state is mapped to the observations, P is the covariance matrix of the state estimation error, representing the uncertainty of the estimation, Q is the process noise covariance matrix, representing the uncertainty of the system model, R is the observation noise covariance matrix, representing the uncertainty of the sensor measurement, and K is the Kalman gain.

[0090] By using Kalman filtering for fusion, sensor data of different accuracies and reliability can be dynamically and optimally fused to obtain smoother and more accurate correlation data.

[0091] Then, using the associated data as input, the corresponding entity objects in the digital twin model are updated with their attributes in real time. This model not only stores static data, but also infers comprehensive state information that cannot be directly observed through a built-in physics engine or business logic.

[0092] In one feasible implementation, after the steps of constructing a dynamic digital twin model describing the operating environment based on the associated data, and fusing the associated data according to the dynamic digital twin model to generate comprehensive status information, the method further includes:

[0093] A probabilistic model is established for the key physical parameters in the dynamic digital twin model, and sensor data is acquired in real time. The key physical parameters are calibrated based on the sensor data to obtain calibration parameters.

[0094] The historical data within a preset time window is compared with the corresponding comprehensive status to determine the model confidence of the probability model;

[0095] When the model confidence level is lower than the confidence threshold, the dynamic digital twin model is reconstructed based on the calibration parameters.

[0096] It should be noted that key physical parameters refer to variables in a digital twin model that significantly impact simulation accuracy and decision-making outcomes, describing the essential properties of the physical world. A probabilistic model is a mathematical model used to represent the uncertainty of variables. It does not provide a single, definitive value for the parameters, but rather describes the possible range of values ​​and their probability distribution. Model confidence is a quantitative indicator used to evaluate the degree of consistency between the predictions of a digital twin model and actual events in the physical world over a period of time. High confidence indicates a reliable model; low confidence indicates a significant bias in the model.

[0097] In the implementation, a probabilistic model is first initialized for the selected key physical parameters. This initial value can be based on the equipment manual or offline calibration. During runtime, the system continuously acquires relevant sensor data, and these observed data will differ from the model's predictions. At this point, the system uses a Bayesian filtering algorithm for parameter calibration. The core of this algorithm is Bayesian update: using the probability distribution of the parameters as a prior and the new observed data as evidence, the updated posterior distribution is calculated using Bayes' theorem. Simultaneously, the system maintains a sliding time window, storing the historical data sequence within this window, including: the model's prediction of the state at time t+1 based on the state at time t, and the actual observed value at time t+1. Confidence is calculated by comparing the difference between the predicted and actual sequences. A common method is to calculate the mean absolute percentage error (MAPE) or the standardized root mean square error (NRMSE) and then map it to a confidence score of 0-1. If the predicted value closely matches the actual value with a small error, the confidence score is close to 1; conversely, if the deviation remains large, the confidence score will gradually decrease.

[0098] The system sets a confidence threshold (e.g., 0.7). When the calculated model confidence consistently falls below this threshold, it indicates that fine-tuning the parameters alone is insufficient to accurately reflect reality. This typically signifies a structural change in the physical system. At this point, the system triggers a model reconstruction process. Using the calibration parameters as a new foundation and incorporating historical operational data, a more accurate and dynamic digital twin model is reconstructed. After reconstruction, the system switches to the new model and resumes monitoring its confidence, forming a complete self-optimization loop.

[0099] Step S20: Input the comprehensive state information into the pre-trained collaborative decision-making model and output the target collaborative strategy that matches the current operation scenario. The collaborative decision-making model is constructed based on deep reinforcement learning.

[0100] It's important to note that the collaborative decision-making model is an artificial intelligence model. Its core function is to comprehensively analyze the state of all entities in the system and generate a globally optimal action plan that coordinates multiple heterogeneous devices to jointly achieve complex goals, rather than making isolated decisions for individual devices. The operational scenario refers to the global state of the system at a specific moment, defined by comprehensive state information. It includes all dynamic and static elements, such as task requirements, device availability, resource distribution, and environmental conditions, and serves as the basis for decision-making. The goal-coordination strategy refers to the set of specific action instructions output by the collaborative decision-making model. It details the specific actions, timing, and coordination methods that each relevant device in the system should perform in the following time period to optimize overall efficiency, cost, or other objectives.

[0101] Understandably, a pre-trained collaborative decision-making model is typically a deep neural network (such as an Actor-Critic architecture) that maps comprehensive state information to a target collaborative strategy. The model is trained on massive operational scenarios simulated by a digital twin model using reinforcement learning algorithms: the model tries different strategies and continuously adjusts its network parameters based on the reward signals generated by the overall system benefits after strategy execution, ultimately learning which collaborative actions to take in which global state to maximize long-term cumulative rewards. In actual deployment, the model performs forward propagation calculations based on real-time comprehensive state information, outputting the current optimal target collaborative strategy, thereby achieving dynamic, intelligent scheduling and collaborative control of the operational site.

[0102] In one feasible implementation, the step of inputting the comprehensive state information into a pre-trained collaborative decision-making model and outputting a target collaborative strategy that matches the current operational scenario includes:

[0103] Define a reward function that aims to optimize task completion efficiency and system resources;

[0104] The comprehensive state information and reward function are input into the pre-trained collaborative decision-making model, so that the pre-trained collaborative decision-making model performs cumulative reward evaluation on collaborative decisions based on the comprehensive state information and the reward function, and obtains a reward evaluation value;

[0105] A probability distribution of actions is generated based on the reward evaluation value, and the target action sequence is determined by the action sequence of the probability distribution of the action sequence corresponding to the maximum reward evaluation value.

[0106] The target action sequence is used as a target collaboration strategy that matches the current operational scenario.

[0107] It's important to note that the reward function quantifies the benefit or penalty gained by a collaborative decision-making model after taking a certain action in a specific state. It is fed back to the model as a reward value and is the sole criterion guiding the model to learn the optimal policy. In this scenario, the function is explicitly designed to encourage high task completion efficiency and optimal system resource utilization. The reward evaluation value is a specific numerical value representing the expected cumulative reward that can be obtained by following a certain policy starting from the current state.

[0108] In the specific implementation, we first define a reward function that aims at improving task completion efficiency and optimizing system resources. The reward function is as follows:

[0109]

[0110] in, For the comprehensive state information at time step t, For the action selected at time step t, A sub-reward function to measure the contribution of an action to the efficiency of task completion. A sub-reward function to measure the contribution of an action to resource optimization. These are the weighting coefficients.

[0111] Then, the Critic network is used to evaluate the state. The expected cumulative reward obtained by following the current strategy is also known as the state value or Q-value. A common method is to discount the cumulative reward:

[0112]

[0113] In practice, The reward is the evaluation value.

[0114] Then, an action probability distribution is generated based on the reward evaluation value. The target action sequence is determined by the action sequence of the action probability distribution corresponding to the maximum reward evaluation value. The target action sequence is then used as the target collaborative strategy that matches the current operational scenario.

[0115] In one feasible implementation, after the step of using the target action sequence as a target collaborative strategy matching the current operational scenario, the method further includes:

[0116] The target cooperative strategy was simulated and verified, and the simulation verification results were obtained.

[0117] If the simulation verification result shows that the execution effect of the target collaborative strategy does not reach the predetermined threshold, the verification is deemed to have failed, and an abnormal scenario is identified.

[0118] The pre-trained collaborative decision-making model is retrained based on the abnormal scenario to obtain an optimized collaborative decision-making model;

[0119] The optimized collaborative decision-making model is then released in a canary release phase, replacing the pre-trained collaborative decision-making model.

[0120] In its implementation, the generated target collaborative strategy is input into a high-fidelity dynamic digital twin model. The digital twin model rigorously simulates the entire strategy execution process and records key performance indicators, such as total task completion time, total energy consumption, and equipment utilization. These indicators constitute the simulation verification results. The system compares the simulation results with predetermined thresholds. If any indicator fails to meet the threshold, the verification is deemed a failure.

[0121] When verification fails, the current overall state information is identified as an anomalous scenario. The system records the scenario data that caused the failure, along with the suboptimal policy output by the model. Simultaneously, the system can use more computationally intensive search algorithms or expert rules within the digital twin environment to find a better policy that reaches a predetermined threshold for this anomalous scenario. The system adds the collected anomalous scenarios and their corresponding optimized policy samples to the model's training dataset. Then, a retraining process is initiated. This process typically employs incremental learning or fine-tuning techniques, training the model with a new dataset based on the pre-trained model's weights. The goal is to adjust the model parameters so that, when faced with similar anomalous scenarios, it can output actions closer to the target collaborative policy, thereby correcting the model's shortcomings. After retraining, an optimized collaborative decision-making model is obtained.

[0122] To ensure system stability and prevent global failures due to unknown defects in the new model, a canary release approach will be adopted. A diagram illustrating a canary release is shown below. Figure 2 As shown. For example, initially route 10% of decision requests to the new model, while the remaining 90% are still handled by the old model. Monitor the decision-making effectiveness of this 10% traffic in real time. If the new model performs stably and is better than or equal to the old model during the gray-scale period, gradually increase the gray-scale ratio until the old model is completely replaced. If the new model encounters problems, quickly roll back to the old model to minimize the impact.

[0123] Step S30: Based on the target collaboration strategy and the comprehensive status information, determine the target device cluster that needs to perform collaborative tasks.

[0124] It should be noted that the target device cluster refers to the optimal subset of devices dynamically selected from the global device pool based on policy requirements and resource constraints, which are capable of collaboratively executing specific tasks.

[0125] Understandably, the process begins by analyzing the action sequences within the target coordination strategy to extract its hard constraints and soft preferences regarding device capabilities, resource status, and spatial relationships. Subsequently, a real-time device map is constructed based on comprehensive status information, and a multi-objective optimization algorithm is used for device-task matching. The final output is a target device cluster that meets the coordination requirements and achieves optimal global efficiency, laying the foundation for subsequent task allocation and execution. This process must balance load balancing, fault tolerance, and real-time performance to ensure the cluster can efficiently adapt to the dynamic requirements of the strategy.

[0126] In one feasible implementation, the step of determining the target device cluster that needs to perform collaborative tasks based on the target collaboration strategy and the comprehensive status information includes:

[0127] Analyze the target collaboration strategy to obtain the core task type and required equipment capability combination corresponding to the target collaboration strategy;

[0128] Based on the combination of device capabilities, a set of candidate devices that meet the criteria is selected from the currently online and available devices;

[0129] Based on the comprehensive status information, the real-time load and geographical location information of the candidate device set are determined, the collaborative communication cost is evaluated, and the target device cluster that needs to perform collaborative tasks is determined based on the real-time load, the geographical location information, and the evaluated collaborative communication cost.

[0130] In its implementation, the system first performs semantic parsing and structured decomposition of the target collaboration strategy, identifies the key capability dimensions required for the task, and performs capability tag matching in the device metadata knowledge base to initially screen out a set of candidate devices that meet the functional requirements. Subsequently, combined with real-time monitoring data streams, the system quantitatively evaluates the dynamic load status and physical deployment location of the candidate devices, and constructs a network topology cost model based on inter-device communication latency and bandwidth constraints. Finally, by introducing a multi-factor weighted evaluation algorithm, the system selects the target device cluster from the candidate set that meets the collaborative task requirements and has the greatest overall collaborative benefit, while reserving redundancy mechanisms for fault switching and elastic expansion.

[0131] Step S40: Generate differentiated control instructions for each device in the target device cluster according to the objective of the collaborative task.

[0132] It should be noted that the goal of a collaborative task refers to the specific quantifiable result or state that the task expects to achieve, such as identifying and reporting the coordinates of all target objects within a region. Control commands refer to the specific sequence of commands that directly drive the underlying execution units of the device to perform operations, containing precise information such as action type, parameters, and timing.

[0133] Understandably, the overall goal of the collaborative task is first decomposed into a series of atomic subtasks based on the role, capabilities, and real-time status of each device in the target device cluster. Then, based on device differences and real-time environment constraints, a set of control instructions that precisely matches each device is generated to ensure that the instructions are synchronized in time and coordinated in logic. Finally, the differentiated control instructions are accurately distributed to the corresponding device execution units through the instruction distribution module, driving the cluster to form a unified and efficient overall collaborative behavior.

[0134] In one feasible implementation, the step of generating differentiated control instructions for each device in the target device cluster based on the objective of the collaborative task includes:

[0135] The collaborative task is decomposed into multiple atomic subtasks with logical order and dependencies.

[0136] Based on the roles and capabilities of each device in the target device cluster, the atomic subtasks are assigned to the corresponding devices;

[0137] For each assigned atomic subtask, a precise start timestamp and timeout threshold are set, and the interaction interface protocol required for data transfer between devices is configured to form the differentiated control commands.

[0138] In its implementation, the system performs structured parsing of the top-level collaborative task, using a pre-defined task tree model or workflow engine to decompose it layer by layer into multiple indivisible, functionally defined atomic subtasks. This process strictly defines the logical relationships between subtasks, such as sequential execution, parallel execution, or conditional dependencies, forming a directed acyclic graph to describe the overall task flow.

[0139] Next, the system intelligently allocates the decomposed atomic subtasks to the most suitable devices in the target cluster based on a predefined or dynamically negotiated "role-capability" mapping relationship. The allocation strategy not only considers whether the core functions of the devices match, but also comprehensively considers their current real-time load, computing power, and expected energy consumption of task execution, in order to achieve load balancing and efficiency optimization.

[0140] Finally, to ensure precise synchronization of the entire cluster's collaborative actions in time, the system calculates an absolute timestamp or relative trigger delay for each subtask as a start command and sets a reasonable timeout threshold for fault detection. Simultaneously, the command generation module pre-configures a unified data format, communication protocol, and destination address for task nodes involved in inter-device data exchange. This ensures that the command not only tells the device what to do but also specifies when to do it and how to collaborate with its peers, thus encapsulating it into a highly differentiated set of control commands that can ultimately be directly parsed and executed by the devices.

[0141] Step S50: Distribute the differentiated control commands to the corresponding target devices to drive the target device cluster to collaboratively complete the collaborative task.

[0142] Understandably, a message middleware or dedicated communication gateway is used to accurately address the device by leveraging the unique network identifier obtained during device registration. High-reliability communication protocols are prioritized to serialize and send differentiated control commands to the corresponding target devices. To address network fluctuations and device offline risks, message queue persistence, command confirmation, and retransmission mechanisms are typically employed to ensure command delivery. After receiving the commands, the agent program on the device verifies and parses them, ultimately calling the device's local SDK or API to drive the hardware execution unit, thereby transforming the collaborative strategy of the decision-making center into collaborative operations of the device cluster in the physical world.

[0143] In one feasible implementation, after the step of distributing the differentiated control commands to the corresponding target devices to drive the target device cluster to collaboratively complete the collaborative task, the method further includes:

[0144] It receives instruction execution feedback data and new environmental status data from the target device cluster in real time.

[0145] The feedback data is compared with the expected execution status. If there is equipment malfunction, execution delay or sudden environmental change, it is determined to be a coordination deviation.

[0146] Based on the type and severity of the coordination deviation, the unexecuted control instructions are dynamically adjusted or the coordination decision-making model is triggered to make a new decision.

[0147] It should be noted that instruction execution feedback data refers to the data about the device's own execution status that the device actively reports after executing the control instruction, while new environmental status data refers to environmental change information that may affect the execution of the task and is newly detected by devices within the cluster or external sensors during the task execution process.

[0148] In the specific implementation, a high-performance message receiving service continuously listens for and aggregates feedback data reported by various devices in the cluster at fixed frequencies or triggered events, while subscribing to the latest status streams published by the environment awareness module. This real-time data is sent to a status monitoring and comparison engine, which compares the actual data with the expected status trajectory generated based on the original instruction sequence in real time.

[0149] When the engine detects a deviation, such as a fault code reported by a device, a significant delay in execution, or a sensor reporting an obstacle not anticipated during planning, it will automatically determine the occurrence of the coordination deviation event according to preset rules and immediately assess its type and severity level.

[0150] Based on the evaluation results, the system will activate different levels of corrective mechanisms:

[0151] For minor, localized deviations, the system will invoke the instruction dynamic adjuster to quickly calculate the correction scheme for subsequent instructions and immediately issue the adjusted instructions.

[0152] For severe or global deviations, it means that the original task planning is no longer applicable. The system will interrupt the current execution flow and send a re-decision request containing the latest global state to the collaborative decision-making model, thereby generating a brand-new task decomposition and instruction set to guide the cluster to adapt to the new situation and ensure the final completion of the collaborative task.

[0153] In one feasible implementation, the step of dynamically adjusting unexecuted control instructions or triggering the collaborative decision-making model to make a re-decision based on the type and severity of the coordination deviation includes:

[0154] If a device fails, the failed device is isolated from the target device cluster, and the unfinished atomic subtasks of the failed device are reassigned to other available devices in the cluster.

[0155] If a delay in execution timing is detected, the delay amount is calculated, and the start time of the unexecuted control instructions is proportionally compressed or delayed based on the delay amount.

[0156] All dynamic adjustment actions, reasons for adjustment, and results of adjustment are recorded in a log, and the collaborative decision-making model is triggered to make a new decision based on the log.

[0157] It's important to note that device isolation is a security and resource management mechanism. When the system confirms that a device has experienced an unrecoverable failure, it is logically removed from the list of active devices. This ensures that subsequent task allocation and cluster collaboration planning no longer consider this device, preventing the sending of invalid commands or waiting for invalid responses. Timing delay refers to the offset between the actual execution time of a task and its originally planned time. This delay may be caused by timeouts of individual devices, network jitter, or delays in the completion of preceding tasks. Proportional compression is a strategy for adjusting the schedules of subsequent tasks. With a fixed total time limit, by uniformly shortening the time intervals of all subsequent unexecuted tasks, it aims to make up for time lost due to earlier delays and strive to complete tasks before the final deadline.

[0158] In practice, when the monitoring system determines a coordination deviation, the scheduler will trigger a preset response strategy based on the type of deviation.

[0159] If the deviation is caused by a device malfunction, the scheduler will first mark the device as offline or faulty in the cluster state management to achieve logical isolation. Then, it will retrieve the task dependency graph, identify all atomic subtasks assigned to the faulty device that have not yet started or completed, and immediately launch a fast reallocation algorithm. This algorithm, based on the real-time status of the remaining available devices, aims to minimize the overall task completion time, reallocates these tasks, and generates new control commands for issuance.

[0160] If the deviation is an execution timing delay, the scheduler will first accurately calculate the accumulated delay. Then, it will analyze the flexibility of the unexecuted task chain: if there is an idle time buffer between subsequent tasks and the final deadline is strict, a proportional compression strategy will be adopted to reduce the time interval of all subsequent tasks in order to make up for the time; if the final deadline is relatively flexible or compression would be risky, a more conservative postponement adjustment will be adopted, pushing the entire subsequent plan back by the delay amount to ensure that the coordination between tasks is not disrupted.

[0161] All these decisions and actions, including the triggered rules, the specific parameters adjusted, and the results of the redistribution, are written to the dynamic adjustment action log in real time and in a structured manner. This log is not only used for operational auditing, but more importantly, it serves as a key input to the upper-level collaborative decision-making model. The model can learn offline or optimize online based on these real-world corrective cases, thereby better anticipating potential problems when planning future tasks, generating more robust initial plans, and gradually improving the intelligence and adaptability of the entire cluster system.

[0162] This embodiment provides a smart operation management method based on multi-device collaboration. It receives real-time operational data from multiple heterogeneous devices, fuses and processes this data to generate comprehensive status information, inputs this comprehensive status information into a pre-trained collaborative decision-making model, and outputs a target collaborative strategy matching the current operational scenario. Based on the target collaborative strategy and the comprehensive status information, it determines the target device cluster that needs to perform collaborative tasks. According to the objective of the collaborative task, it generates differentiated control instructions for each device in the target device cluster and distributes these differentiated control instructions to the corresponding target devices to drive the target device cluster to collaboratively complete the collaborative task. This method overcomes the technical problem of low collaborative efficiency caused by rigid decision-making during multi-heterogeneous device collaborative operation.

[0163] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the intelligent operation management method based on multi-device collaboration in this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0164] This application also provides a smart operation management system based on multi-device collaboration, please refer to... Figure 3 The intelligent operation management system based on multi-device collaboration includes:

[0165] The data fusion module 10 is used to receive real-time operating data from multiple heterogeneous devices, and to perform fusion processing on the real-time operating data to generate comprehensive status information;

[0166] The intelligent collaboration module 20 is used to input the comprehensive state information into a pre-trained collaborative decision-making model and output a target collaborative strategy that matches the current operation scenario. The collaborative decision-making model is built based on deep reinforcement learning.

[0167] The device determination module 30 is used to determine the target device cluster that needs to perform the collaborative task based on the target collaboration strategy and the comprehensive status information.

[0168] The instruction generation module 40 is used to generate differentiated control instructions for each device in the target device cluster according to the objective of the collaborative task.

[0169] The dynamic control module 50 is used to distribute the differentiated control commands to the corresponding target devices to drive the target device cluster to collaboratively complete the collaborative task.

[0170] In one feasible implementation, the data fusion module 10 is further configured to receive real-time operating data from multiple heterogeneous devices, perform data cleaning and format standardization on the received real-time operating data from each device, and obtain standardized data.

[0171] Based on a multi-source information fusion algorithm, data from different devices but describing the same operating object in the standardized data are correlated and complemented to obtain correlated data;

[0172] A dynamic digital twin model describing the operating environment is constructed from the associated data. The associated data is then fused and processed based on the dynamic digital twin model to generate comprehensive status information.

[0173] In one feasible implementation, the data fusion module 10 is further configured to establish a probabilistic model for the key physical parameters in the dynamic digital twin model, acquire sensor data in real time, and perform parameter calibration on the key physical parameters based on the sensor data to obtain calibration parameters.

[0174] The historical data within a preset time window is compared with the corresponding comprehensive status to determine the model confidence of the probability model;

[0175] When the model confidence level is lower than the confidence threshold, the dynamic digital twin model is reconstructed based on the calibration parameters.

[0176] In one feasible implementation, the intelligent collaboration module 20 is further used to define a reward function with the goal of optimizing task completion efficiency and system resources;

[0177] The comprehensive state information and reward function are input into the pre-trained collaborative decision-making model, so that the pre-trained collaborative decision-making model performs cumulative reward evaluation on collaborative decisions based on the comprehensive state information and the reward function, and obtains a reward evaluation value;

[0178] A probability distribution of actions is generated based on the reward evaluation value, and the target action sequence is determined by the action sequence of the probability distribution of the action sequence corresponding to the maximum reward evaluation value.

[0179] The target action sequence is used as a target collaboration strategy that matches the current operational scenario.

[0180] In one feasible implementation, the intelligent collaboration module 20 is further used to simulate and verify the target collaboration strategy to obtain simulation and verification results;

[0181] If the simulation verification result shows that the execution effect of the target collaborative strategy does not reach the predetermined threshold, the verification is deemed to have failed, and an abnormal scenario is identified.

[0182] The pre-trained collaborative decision-making model is retrained based on the abnormal scenario to obtain an optimized collaborative decision-making model;

[0183] The optimized collaborative decision-making model is then released in a canary release phase, replacing the pre-trained collaborative decision-making model.

[0184] In one feasible implementation, the device determination module 30 is further configured to parse the target collaboration strategy and obtain the core task type and required device capability combination corresponding to the target collaboration strategy;

[0185] Based on the combination of device capabilities, a set of candidate devices that meet the criteria is selected from the currently online and available devices;

[0186] Based on the comprehensive status information, the real-time load and geographical location information of the candidate device set are determined, the collaborative communication cost is evaluated, and the target device cluster that needs to perform collaborative tasks is determined based on the real-time load, the geographical location information, and the evaluated collaborative communication cost.

[0187] In one feasible implementation, the instruction generation module 40 is further configured to decompose the collaborative task into multiple atomic subtasks with logical order and dependency relationship.

[0188] Based on the roles and capabilities of each device in the target device cluster, the atomic subtasks are assigned to the corresponding devices;

[0189] For each assigned atomic subtask, a precise start timestamp and timeout threshold are set, and the interaction interface protocol required for data transfer between devices is configured to form the differentiated control commands.

[0190] In one feasible implementation, the dynamic control module 50 is also used to receive instruction execution feedback data and new environmental status data from the target device cluster in real time.

[0191] The feedback data is compared with the expected execution status. If there is equipment malfunction, execution delay or sudden environmental change, it is determined to be a coordination deviation.

[0192] Based on the type and severity of the coordination deviation, the unexecuted control instructions are dynamically adjusted or the coordination decision-making model is triggered to make a new decision.

[0193] In one feasible implementation, the dynamic control module 50 is further configured to isolate the faulty device from the target device cluster if a device malfunctions, and to reassign the unfinished atomic subtasks of the faulty device to other available devices within the cluster.

[0194] If a delay in execution timing is detected, the delay amount is calculated, and the start time of the unexecuted control instructions is proportionally compressed or delayed based on the delay amount.

[0195] All dynamic adjustment actions, reasons for adjustment, and results of adjustment are recorded in a log, and the collaborative decision-making model is triggered to make a new decision based on the log.

[0196] The intelligent operation management system based on multi-device collaboration provided in this application, employing the intelligent operation management method based on multi-device collaboration in the above embodiments, can solve the technical problem of rigid decision-making and low collaboration efficiency when multiple heterogeneous devices are operating collaboratively. Compared with the prior art, the beneficial effects of the intelligent operation management system based on multi-device collaboration provided in this application are the same as those of the intelligent operation management method based on multi-device collaboration provided in the above embodiments, and other technical features in the intelligent operation management system based on multi-device collaboration are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0197] This application provides a smart operation management device based on multi-device collaboration. The smart operation management device based on multi-device collaboration includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the smart operation management method based on multi-device collaboration in the above embodiment 1.

[0198] The following is for reference. Figure 4This document illustrates a structural diagram of a smart operation management device based on multi-device collaboration suitable for implementing embodiments of this application. The smart operation management device based on multi-device collaboration in these embodiments may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 4 The intelligent operation management device based on multi-device collaboration shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0199] like Figure 4 As shown, the intelligent operation management device based on multi-device collaboration may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to programs stored in ROM (Read Only Memory) 1002 or programs loaded from storage device 1003 into RAM (Random Access Memory) 1004. RAM 1004 also stores various programs and data required for the operation of the intelligent operation management device based on multi-device collaboration. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via bus 1005. Input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, LCDs (Liquid Crystal Displays), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows the intelligent operation management device based on multi-device collaboration to exchange data with other devices wirelessly or via wired communication. Although the figure shows an intelligent operation management device based on multi-device collaboration with various systems, it should be understood that it is not required to implement or possess all the systems shown. More or fewer systems can be implemented alternatively.

[0200] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication system, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0201] The intelligent operation management device based on multi-device collaboration provided in this application, employing the intelligent operation management method based on multi-device collaboration in the above embodiments, can solve the technical problems of intelligent operation management based on multi-device collaboration. Compared with the prior art, the beneficial effects of the intelligent operation management device based on multi-device collaboration provided in this application are the same as the beneficial effects of the intelligent operation management method based on multi-device collaboration provided in the above embodiments, and other technical features in this intelligent operation management device based on multi-device collaboration are the same as those disclosed in the method of the previous embodiment, and will not be repeated here.

[0202] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0203] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0204] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, which are used to execute the intelligent operation management method based on multi-device collaboration in the above embodiments.

[0205] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, RAM (Random Access Memory), ROM (Read Only Memory), EPROM (Erasable Programmable Read Only Memory or Flash Memory), optical fibers, CD-ROM (CD-Read Only Memory), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0206] The aforementioned computer-readable storage medium may be included in a smart operation and management device based on multi-device collaboration; or it may exist independently and not be assembled into a smart operation and management device based on multi-device collaboration.

[0207] The aforementioned computer-readable storage medium carries one or more programs. When these programs are executed by a smart operation and management device based on multi-device collaboration, the smart operation and management device based on multi-device collaboration enables the following: receiving real-time operating data from multiple heterogeneous devices and performing fusion processing on the real-time operating data to generate comprehensive status information; inputting the comprehensive status information into a pre-trained collaborative decision model and outputting a target collaborative strategy matching the current operating scenario, wherein the collaborative decision model is constructed based on deep reinforcement learning; determining the target device cluster that needs to perform collaborative tasks based on the target collaborative strategy and the comprehensive status information; generating differentiated control instructions for each device in the target device cluster according to the objective of the collaborative task; and distributing the differentiated control instructions to the corresponding target devices to drive the target device cluster to collaboratively complete the collaborative task.

[0208] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including LAN (Local Area Network) or WAN (Wide Area Network)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0209] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0210] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0211] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described intelligent operation management method based on multi-device collaboration, and can solve the technical problems of intelligent operation management based on multi-device collaboration. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as the beneficial effects of the intelligent operation management method based on multi-device collaboration provided in the above embodiments, and will not be repeated here.

[0212] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the intelligent operation management method based on multi-device collaboration as described above.

[0213] The computer program product provided in this application can solve the technical problem of intelligent operation management based on multi-device collaboration. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as the beneficial effects of the intelligent operation management method based on multi-device collaboration provided in the above embodiments, and will not be repeated here.

[0214] The above are only some embodiments of this application and do not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the content of this application specification and drawings, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A smart operation management method based on multi-device collaboration, characterized in that, The intelligent operation management method based on multi-device collaboration includes: Receive real-time operating data from multiple heterogeneous devices, and perform fusion processing on the real-time operating data to generate comprehensive status information; The comprehensive state information is input into a pre-trained collaborative decision-making model, which outputs a target collaborative strategy that matches the current operational scenario. The collaborative decision-making model is built based on deep reinforcement learning. Based on the target collaboration strategy and the comprehensive status information, the target device cluster that needs to perform collaborative tasks is determined; Based on the objective of the collaborative task, generate differentiated control instructions for each device in the target device cluster; The differentiated control commands are distributed to the corresponding target devices to drive the target device cluster to collaboratively complete the collaborative task; The step of receiving real-time operating data from multiple heterogeneous devices and fusing the real-time operating data to generate comprehensive status information includes: Receive real-time operating data from multiple heterogeneous devices, perform data cleaning and format standardization on the received real-time operating data from each device, and obtain standardized data; Based on a multi-source information fusion algorithm, data from different devices but describing the same operating object in the standardized data are correlated and complemented to obtain correlated data; A dynamic digital twin model describing the operating environment is constructed from the associated data. The associated data is then fused based on the dynamic digital twin model to generate comprehensive status information. The step of inputting the comprehensive state information into the pre-trained collaborative decision-making model and outputting a target collaborative strategy that matches the current operational scenario includes: Define a reward function that aims to optimize task completion efficiency and system resources; The comprehensive state information and reward function are input into the pre-trained collaborative decision-making model, so that the pre-trained collaborative decision-making model performs cumulative reward evaluation on collaborative decisions based on the comprehensive state information and the reward function, and obtains a reward evaluation value; A probability distribution of actions is generated based on the reward evaluation value, and the target action sequence is determined by the action sequence of the probability distribution of the action sequence corresponding to the maximum reward evaluation value. The target action sequence is used as a target collaboration strategy that matches the current operational scenario. After the step of using the target action sequence as a target collaborative strategy matching the current operational scenario, the method further includes: The target cooperative strategy was simulated and verified, and the simulation verification results were obtained. If the simulation verification result shows that the execution effect of the target collaborative strategy does not reach the predetermined threshold, the verification is deemed to have failed, and an abnormal scenario is identified. The pre-trained collaborative decision-making model is retrained based on the abnormal scenario to obtain an optimized collaborative decision-making model; The optimized collaborative decision-making model is then released in a canary release phase, replacing the pre-trained collaborative decision-making model.

2. The method as described in claim 1, characterized in that, After the steps of constructing a dynamic digital twin model describing the operating environment based on the associated data, and fusing the associated data according to the dynamic digital twin model to generate comprehensive status information, the method further includes: A probabilistic model is established for the key physical parameters in the dynamic digital twin model, and sensor data is acquired in real time. The key physical parameters are calibrated based on the sensor data to obtain calibration parameters. The historical data within a preset time window is compared with the corresponding comprehensive status to determine the model confidence of the probability model; When the model confidence level is lower than the confidence threshold, the dynamic digital twin model is reconstructed based on the calibration parameters.

3. The method as described in claim 1, characterized in that, The step of determining the target device cluster that needs to perform collaborative tasks based on the target collaboration strategy and the comprehensive status information includes: Analyze the target collaboration strategy to obtain the core task type and required equipment capability combination corresponding to the target collaboration strategy; Based on the combination of device capabilities, a set of candidate devices that meet the criteria is selected from the currently online and available devices; Based on the comprehensive status information, the real-time load and geographical location information of the candidate device set are determined, the collaborative communication cost is evaluated, and the target device cluster that needs to perform collaborative tasks is determined based on the real-time load, the geographical location information, and the evaluated collaborative communication cost.

4. The method as described in claim 1, characterized in that, The step of generating differentiated control instructions for each device in the target device cluster based on the objective of the collaborative task includes: The collaborative task is decomposed into multiple atomic subtasks with logical order and dependencies. Based on the roles and capabilities of each device in the target device cluster, the atomic subtasks are assigned to the corresponding devices; For each assigned atomic subtask, a precise start timestamp and timeout threshold are set, and the interaction interface protocol required for data transfer between devices is configured to form the differentiated control commands.

5. The method as described in claim 1, characterized in that, After the step of distributing the differentiated control commands to the corresponding target devices to drive the target device cluster to collaboratively complete the collaborative task, the method further includes: It receives instruction execution feedback data and new environmental status data from the target device cluster in real time. The feedback data is compared with the expected execution status. If there is equipment malfunction, execution delay or sudden environmental change, it is determined to be a coordination deviation. Based on the type and severity of the coordination deviation, the unexecuted control instructions are dynamically adjusted or the coordination decision-making model is triggered to make a new decision.

6. The method as described in claim 5, characterized in that, The step of dynamically adjusting unexecuted control instructions or triggering the collaborative decision-making model to make a re-decision based on the type and severity of the coordination deviation includes: If a device fails, the failed device is isolated from the target device cluster, and the unfinished atomic subtasks of the failed device are reassigned to other available devices in the cluster. If a delay in execution timing is detected, the delay amount is calculated, and the start time of the unexecuted control instructions is proportionally compressed or delayed based on the delay amount. All dynamic adjustment actions, reasons for adjustment, and results of adjustment are recorded in a log, and the collaborative decision-making model is triggered to make a new decision based on the log.

7. A smart operation management system based on multi-device collaboration, characterized in that, The intelligent operation management system based on multi-device collaboration includes: The data fusion module is used to receive real-time operating data from multiple heterogeneous devices, and to fuse the real-time operating data to generate comprehensive status information. The intelligent collaboration module is used to input the comprehensive state information into a pre-trained collaborative decision-making model and output a target collaboration strategy that matches the current operation scenario. The collaborative decision-making model is built based on deep reinforcement learning. The device determination module is used to determine the target device cluster that needs to perform the collaborative task based on the target collaboration strategy and the comprehensive status information. The instruction generation module is used to generate differentiated control instructions for each device in the target device cluster according to the objective of the collaborative task. The dynamic control module is used to distribute the differentiated control commands to the corresponding target devices to drive the target device cluster to collaboratively complete the collaborative task. The step of receiving real-time operating data from multiple heterogeneous devices and fusing the real-time operating data to generate comprehensive status information includes: Receive real-time operating data from multiple heterogeneous devices, perform data cleaning and format standardization on the received real-time operating data from each device, and obtain standardized data; Based on a multi-source information fusion algorithm, data from different devices but describing the same operating object in the standardized data are correlated and complemented to obtain correlated data; A dynamic digital twin model describing the operating environment is constructed from the associated data. The associated data is then fused based on the dynamic digital twin model to generate comprehensive status information. The step of inputting the comprehensive state information into the pre-trained collaborative decision-making model and outputting a target collaborative strategy that matches the current operational scenario includes: Define a reward function that aims to optimize task completion efficiency and system resources; The comprehensive state information and reward function are input into the pre-trained collaborative decision-making model, so that the pre-trained collaborative decision-making model performs cumulative reward evaluation on collaborative decisions based on the comprehensive state information and the reward function, and obtains a reward evaluation value; A probability distribution of actions is generated based on the reward evaluation value, and the target action sequence is determined by the action sequence of the probability distribution of the action sequence corresponding to the maximum reward evaluation value. The target action sequence is used as a target collaboration strategy that matches the current operational scenario. After the step of using the target action sequence as a target collaborative strategy matching the current operational scenario, the method further includes: The target cooperative strategy was simulated and verified, and the simulation verification results were obtained. If the simulation verification result shows that the execution effect of the target collaborative strategy does not reach the predetermined threshold, the verification is deemed to have failed, and an abnormal scenario is identified. The pre-trained collaborative decision-making model is retrained based on the abnormal scenario to obtain an optimized collaborative decision-making model; The optimized collaborative decision-making model is then released in a canary release phase, replacing the pre-trained collaborative decision-making model.

Images